Dentomaxillofacial Radiology (2016) 45, 20150240 ª 2015 The Authors. Published by the British Institute of Radiology birpublications.org/dmfr
RESEARCH ARTICLE
Quantitative and qualitative comparison of MR imaging of the temporomandibular joint at 1.5 and 3.0 T using an optimized high-resolution protocol 1,2 1
Andrei Manoliu, 2Georg Spinner, 2Michael Wyss, 3Stefan Erni, 3Dominik A Ettlin, 1Daniel Nanz, Erika J Ulbrich, 3Luigi M Gallo and 1Gustav Andreisek
1 Department of Radiology, Institute for Diagnostic and Interventional Radiology, University Hospital Zurich, University of Zurich, Zurich, Switzerland; 2Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zurich, Switzerland; 3 Center of Dental Medicine of the University of Zurich, Zurich, Switzerland
Objectives: To quantitatively and qualitatively compare MRI of the temporomandibular joint (TMJ) using an optimized high-resolution protocol at 3.0 T and a clinical standard protocol at 1.5 T. Methods: A phantom and 12 asymptomatic volunteers were MR imaged using a 2-channel surface coil (standard TMJ coil) at 1.5 and 3.0 T (Philips Achieva and Philips Ingenia, respectively; Philips Healthcare, Best, Netherlands). Imaging protocol consisted of coronal and oblique sagittal proton density-weighted turbo spin echo sequences. For quantitative evaluation, a spherical phantom was imaged. Signal-to-noise ratio (SNR) maps were calculated on a voxelwise basis. For qualitative evaluation, all volunteers underwent MRI of the TMJ with the jaw in closed position. Two readers independently assessed visibility and delineation of anatomical structures of the TMJ and overall image quality on a 5-point Likert scale. Quantitative and qualitative measurements were compared between field strengths. Results: The quantitative analysis showed similar SNR for the high-resolution protocol at 3.0 T compared with the clinical protocol at 1.5 T. The qualitative analysis showed significantly better visibility and delineation of clinically relevant anatomical structures of the TMJ, including the TMJ disc and pterygoid muscle as well as better overall image quality at 3.0 T than at 1.5 T. Conclusions: The presented results indicate that expected gains in SNR at 3.0 T can be used to increase the spatial resolution when imaging the TMJ, which translates into increased visibility and delineation of anatomical structures of the TMJ. Therefore, imaging at 3.0 T should be preferred over 1.5 T for imaging the TMJ. Dentomaxillofacial Radiology (2016) 45, 20150240. doi: 10.1259/dmfr.20150240 Cite this article as: Manoliu A, Spinner G, Wyss M, Erni S, Ettlin DA, Nanz D, et al. Quantitative and qualitative comparison of MR imaging of the temporomandibular joint at 1.5 and 3.0 T using an optimized high-resolution protocol. Dentomaxillofac Radiol 2016; 45: 20150240. Keywords: temporomandibular joint; temporomandibular joint disc; temporomandibular joint disorders; magnetic resonance imaging (MRI); signal-to-noise ratio (SNR)
Introduction Temporomandibular disorders (TMDs) are a collective term for various pathologies of the temporomandibular Correspondence to: Andrei Manoliu. E-mail: [email protected] The study was funded by the Swiss National Science Foundation grant number 320030_156466. Received 18 July 2015; revised 27 August 2015; accepted 9 September 2015
joint (TMJ) characterized by mutual features, such as pain, clicking or crepitus of the TMJ and alterations of the mouth opening path.1,2 With a prevalence of 12% in the adult population,3 TMDs are the most common cause for orofacial pain of non-dental origin.4 TMDs have a great impact on the quality of life of patients and
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
2 of 7
are related to a wide diversity of frequent disorders, including tension and migraine headache, depression, fatigue, sleep apnoea, obesity and Type 2 diabetes mellitus.5,6 In addition, TMDs cause great socioeconomic costs, causing 17,800,000 lost workdays for every 100,000,000 working adults per year in the USA.3,7 While the need for early detection of the pathological processes underlying TMD is evident, the diagnostic process is still subject of substantial discussion.8 In cases where imaging is required, MRI has been considered the imaging method of choice over the last decade to evaluate key features of TMD, such as temporomandibular disc displacement or alterations of cartilage and bones.9,10 Currently, MRI of the TMJ is mainly performed at 1.5 T using dedicated surface coils.11 Since the signal intensity [or, in particular, the signal-to-noise ratio (SNR)] is assumed to scale linearly with the field strength, theoretical considerations suggest that imaging the TMJ at higher field strengths such as 3.0 T should allow considerably higher spatial resolutions at comparable scan durations and therefore enable superior visibility of TMJ’s anatomical key structures.12 Although theoretical considerations suggest great advantages for MRI of the TMJ at higher magnetic field strengths, evidence in the literature for the use of 3.0 T is still scarce and only few studies13,14 investigated the benefit of imaging the TMJ at 3.0 T compared with 1.5 T. Stehling et al13 showed a better visibility of some anatomical structures, and Schmid-Schwap et al14 found a better diagnostic accuracy in patients with anterior disc displacement. However, these early studies had several limitations and only used commercially nonavailable custom-made bilateral phased-array surface coils or applied imaging protocols, which have not been optimized for imaging the TMJ at 3.0 T.13,14 In addition, the two aforementioned studies focused on qualitative assessment only, whereas the SNR and its potential impact on imaging results have not been quantitatively assessed so far. Therefore, clear evidence for a superior quantitative and qualitative performance of imaging the TMJ at 3.0 T using commercially available coils and easily reproducible dedicated imaging protocols is still missing and needs to be addressed. The aim of the present study was to quantitatively and qualitatively compare an optimized high-resolution protocol at 3.0 T and a clinical standard protocol at 1.5 T for imaging the TMJ using a commercially available two-channel surface coil. Methods and materials This was a prospective institutional review boardapproved MRI study. It included imaging of a phantom and a cohort of asymptomatic healthy volunteers. All volunteers gave written informed consent. The study was registered in the official research database of the University of Zurich. Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
Phantom imaging The first part of the study included standardized SNR measurements for which a “Braino” phantom (GE Medical Systems, Milwaukee, WI) containing 12.5 mM of N-acetyl-L-aspartic acid (NAA), 10 mM of creatine hydrate (Cr), 3 mM of choline chloride (Ch), 7.5 mM of myoinositol (mI), 12.5 mM of L-glutamic acid (Glu), 5 mM of DL-lactic acid (Lac), sodium azide (0.1%), 50 mM of potassium phosphate monobasic (KH2PO4), 56 mM of sodium hydroxide (NaOH) and 1 ml l21 of Gd-DPTA (Magnevist®; Bayer Healthcare Germany, Leverkusen, Germany) was used. The phantom was a model for the head of a human subject and was placed accordingly on the scanner’s patient table (Figure 1). Volunteer imaging For the second part of the study, 12 asymptomatic healthy volunteers (six females; mean age, 23.16 years; age range, 19–26 years; and six males; mean age, 25.33 years; age range, 20–29 years) who met the following inclusion criteria were included consecutively: age between 18 and 40 years and no past or current TMD. Exclusion criteria were pregnancy, metallic implants (including retainer) or any other contraindications for MRI (e.g. claustrophobia), as well as pain or functional symptoms in the area of the TMJ. All volunteers underwent simultaneous bilateral MRI of their TMJs with the jaw in closed position using the two-channel SENSE Flex S coils (Philips Healthcare, Best, Netherlands) of both scanners (Figure 2). All subjects were positioned in the supine and head-first position. Scans were performed consecutively. To avoid potential imaging bias, half of the volunteers were first assessed at 1.5 T and the other half were assessed at 3.0 T first. The assignment was made randomly. One of the investigators (blinded, AM) was present at all imaging sessions to ensure the correct positioning of the human subjects and coils. Both TMJs were imaged at the same time. MRI MRI was performed on a clinical 1.5-T Philips Achieva system and a clinical 3.0-T Philips Ingenia system (Philips Healthcare) using standard commercially available two-channel surface coils (SENSE Flex S; Philips Healthcare). The phantom and human subjects were imaged using the same hardware and software as well as the same imaging set-up. In general, sequences from the clinical standard protocol were applied. This included proton densityweighted turbo spin echo (PDw-TSE) MR sequences in oblique sagittal plane (field of view) positioned perpendicular to the mandibular condyle’s transverse axis as well as a PDw-TSE MR sequence in coronal plane. For imaging at 1.5 T, sequence parameters applied in clinical routine were used (Table 1). For imaging at 3.0 T, the PDw-TSE sequences were optimized by improving the in-plane resolution of the images (Table 1). Care was taken to keep the acquisition time similar and to not extend the scan duration. This study design was chosen
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
to reflect the clinical situation where 3.0-T protocols are typically different from 1.5 T to take advantage of the higher field strength. To allow standardized SNR measurements in the phantom part of the study, the clinical sequence was accompanied by an additional identical scan without radiofrequency excitation and without gradient switching. Image analysis Quantitative signal-to-noise ratio measurements in the phantom: According to a previously described method12 and strictly following a recently presented analysis algorithm,15 the SNR was examined on a voxelwise basis employing a so-called noise scan, which is basically an additional sequence where the radiofrequency excitation and gradients are switched off by applying the following equation using dedicated in-house software routines (MATLAB®; MathWorks®, Natick, MA): SNR5
jrj s
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dH c 2 1d
where r is the exact magnitude of the available transverse magnetization for each voxel according to Roemer et al,16 s is the standard deviation of its corresponding noise components as described by Pruessmann et al,17 d represents the complex values that a given pixel exhibits in each single-channel image, H denotes the complex conjugate transpose and c denotes the noise covariance matrix calculated from the noise scans as previously described.12,17 Therefore, the applied software routine accounted for possible noise correlations among all coil channels. Subsequently, the phantom was manually segmented by one author (GS), and SNR values were extracted, resulting in one SNR value for each field strength. Qualitative image evaluation in human subjects: All images were transferred to the University Hospital of Zurich’s picture archiving and communication system (IMPAX® 6.0; Agfa Healthcare, Mortsel, Belgium) and independently evaluated by two fellowship-trained radiologists (GA and AM; 14 and 4 years’ experience, respectively, with MRI of the musculoskeletal system). Both readers were blinded with regard to the subject’s details as well as field strength. In accordance to a previously reported grading system,13 images were evaluated in a random order for the overall image quality as well as the visibility and delineation of clinically relevant structures, i.e. the (a) TMJ disc (anterior band, intermediate zone, posterior band), (b) the bilaminar zone, (c) the mandibular fossa, (d) the mandibular condyle and (e) the pterygoid muscle on a 5-point Likert scale from 1 to 5 [1, excellent visibility; 2, good visibility; 3, moderate visibility; 4, poor visibility; and 5, not visible (complete lack of diagnostic information)]. Statistical analyses Two-sample t-tests were used to evaluate potential SNR differences between the images assessed in the phantom
3 of 7
at 1.5 and 3.0 T. Kappa statistics were used to determine the interreader agreement in the qualitative MR image analysis of human subjects. According to Landis and Koch,18 kappa values of 0.41–0.60 were considered as moderate agreement, values of 0.61–0.80 were considered as substantial agreement, values of 0.81–0.99 were considered as almost perfect agreement and values of 1.00 were considered as perfect agreement. To investigate statistically significant between-group differences regarding the visibility and delineation of clinically relevant anatomical structures as well as regarding the overall image quality using different field strengths, Wilcoxon signedrank tests were performed [a-level of 0.05, corrected for multiple comparisons (n 5 13)]. SPSS® software v. 22.0 (IBM Corporation, Armonk, NY; formerly SPSS Inc., Chicago, IL) was used for all statistical analyses. Results Quantitative signal-to-noise ratio data from phantom imaging Quantitative analysis revealed similar SNR for the optimized protocol at 3.0 T compared with the clinical standard protocol at 1.5 T without statistically significant differences: mean ± standard deviation: 1.5 T, 60.42 ± 9.28; 3.0 T, 56.48 ± 7.62; t 5 1.604; p 5 0.116 (Figure 1). Qualitative data from volunteer imaging 24 TMJ image sets of 12 healthy volunteers were evaluated. Interrater reliability ranged from “substantial agreement” to “almost perfect agreement” (Cohen’s kappa: 1.5 T, 0.647–0.934; 3.0 T, 0.739–0.932) for the different anatomical structures (Table 2). For both readers, qualitative analysis demonstrated significantly better overall image quality at 3.0 T than at 1.5 T (Wilcoxon signed-rank test: Reader 1, p 5 0.002; Reader 2, p 5 0.001). Furthermore, imaging at 3.0 T yielded significantly better visibility and delineation of the anterioposterior band of the temporomandibular disc as well as better visibility and delineation of the pterygoid muscle for both readers (Wilcoxon signed-rank tests, p , 0.05, corrected for multiple comparisons, Table 3 and Figure 2). In addition, imaging at 3.0 T yielded better visibility of the intermediate zone of the temporomandibular disc and better visibility of the bilaminar zone; however, statistical significance did not survive correction for multiple comparisons (Table 3). Discussion To the best of our knowledge, the present study demonstrated for the first time that imaging the TMJ at 3.0 T at high resolution yielded better quantitative and qualitative performance than at 1.5 T when using commercially available TMJ surface coils. Despite higher resolution at 3.0 T, SNR was similar, with both magnetic field strengths, suggesting that gains in SNR due to a higher magnetic field birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
4 of 7
Figure 1 Quantitative analysis. Panel (a) shows the spherical phantom. Two stacks consisting of 12 slices each were imaged in sagittal orientation resulting in 24 slices numbered consecutively from left to right. Panel (b) shows a diagram displaying the signal-to-noise ratio (SNR) for both 1.5 and 3.0 T in the spherical phantom. The x-axis represents the number of slices and the y-axis represents the SNR. Values for 1.5 T are given in red and values for the 3.0 T are given in blue. For each slice, the mean SNR is given for both field strengths. Panels (c) and (d) show the voxelwise distribution of SNR for each slice, respectively, whereas panel (c) represents 1.5 T and panel (d) 3.0 T. In both panels (c, d), SNR values are colourcoded from 0 (blue) to 120 (yellow). For colour image see online.
strength can be used to increase spatial resolution while maintaining a robust SNR. Quantitative analysis To the best of our knowledge, no study so far has investigated the SNR at different field strengths for imaging the TMJ. In the present study, we applied an intricate voxelwise approach to calculate SNR for different field strengths in a phantom.15 This approach yielded a comprehensive assessment of the spatial distribution of the SNR (Figure 1) with respect to the desired regions of assessment. Furthermore, our method yielded SNR maps for the whole field of view and correctly accounted for Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
possible noise correlations among all coil channels. Applying this method, we found that SNR was similar for the high-resolution protocol at 3.0 T and the clinical standard protocol at 1.5 T. This result is well in line with the fact that measured SNR is not only related to the magnetic field strength but also to the voxel size.12 In particular, SNR is negatively correlated with the spatial resolution; a higher spatial resolution (i.e. a smaller voxel size) is assumed to lower the SNR. In the present study, the gain in SNR due to the increased field strength at 3.0 T was used to increase the spatial resolution, yielding statistically similar SNR maps for both field strengths while the spatial resolution at 3.0 T was considerably higher
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
Figure 2 Qualitative analysis. proton density-weighted turbo spin echo oblique sagittal images in closed-mouth position at 3.0 T. Panels (a) and (b) show the images of a temporomandibular joint (TMJ) of an asymptomatic volunteer assessed at 1.5 T [panel (a), “good” overall image quality] and 3.0 T [panel (b), “excellent” image quality] using a dedicated, commercially available TMJ surface coil (FlexS; Philips Healthcare). In general, the overall image quality was significantly better on images obtained at 3.0 T. In particular, the visibility and delineation of the anterior band and the posterior band is increased at 3.0 T. See Table 3 for detailed presentation of between-group differences regarding visibility and delineation of all anatomical structures of the TMJ.
than at 1.5 T. Particularly when considering that in many cases the correlation between reported clinical symptoms and MRI findings in patients with TMDs is still unsatisfying,19 exploiting SNR gains at higher field strengths to increase the spatial resolution should be desirable from a clinical perspective, since a better and more detailed depiction of the TMJ’s anatomical subregions is assumed to improve diagnostic sensitivity when evaluating potential pathologies underlying TMDs. Qualitative analysis Using a high-resolution protocol, imaging the TMJ at 3.0 T yielded better visibility and delineation for several key structures of the TMJ, including the temporomandibular disc and the pterygoid muscle. Furthermore, overall image quality was higher at 3.0 T than at 1.5 T. Statistical differences remained significant after correction for multiple
5 of 7
comparisons. It is to note, however, that the depiction of the temporomandibular fossa and condyle did not benefit from increased field strength, which is however clinically less important, since they can be already well depicted at 1.5 T. These findings are in line with previous preliminary studies,13,14 reporting better visibility of the temporomandibular disc at 3.0 T than at 1.5 T. In particular, Stehling et al13 reported increased visibility of the temporomandibular disc in asymptomatic volunteers when imaged at 3.0 T compared with 1.5 T. However, a dedicated custommade phased-array surface coil was used for imaging at 3.0 T, whereas a commercially available surface coil was used at 1.5 T. In another study, Schmidt-Schwap et al14 found that imaging the TMJ at 3.0 T yielded better visibility of the position and shape of the disc and therefore provides increased diagnostic accuracy compared with 1.5 T in patients with suspected anterior disc displacement. However, the sequence parameters at both field strengths were kept almost identical, thus not exploiting potential benefits of higher magnetic field strengths, such as increased spatial resolution. Furthermore, both studies13,14 focused on the temporomandibular disc and condyle only, neglecting other anatomical structures of high relevance regarding TMDs, such as the pterygoid muscle. Therefore, our findings extend the current knowledge as follows. (a) In addition to better visibility and delineation of the temporomandibular disc, we found better visibility and delineation of the pterygoid muscle, particularly the part near the condylar insertion. These findings are of essential clinical interest, since the pathophysiology of TMDs is highly complex and can only be accurately assessed when evaluating the complete musculoskeletal functional unit of the TMJ, including the masticatory muscles. (b) In contrast to previous studies,13 only commercially available coils (FlexS; Philips Healthcare) were used for imaging at both field strengths, ensuring that the reported results are not explained by different array designs. Moreover, the current results can be translated directly into a clinical setting, since the used
Table 1 Scan parameters of the PDw sequences at 1.5 and 3.0 T
Parameter FoV (mm) Pixel size (mm) Reconstructed pixel size (mm) Slice thickness (mm) Number of slices TR TE Wfs in pixel (mm) Effective wfs in image (mm) Wfs (Hz) TSE factor ES NSA Scan time (min)
PDw-TSE sagittal 1.5 T 150 3 150 0.60 3 0.75 0.375 3 0.375 2 2 3 12 2000 30 0.937 0.6 231 7 7.5 2 03 : 48
3.0 T 150 3 150 0.50 3 0.50 0.250 3 0.250 2 2 3 12 2700 26 1.199 0.6 362 7 7.4 1 03 : 52
PDw-TSE coronal 1.5 T 120 3 180 0.60 3 0.75 0.375 3 0.375 2 24 2000 30 0.99 0.6 217 8 8 2 4 : 04
3.0 T 120 3 180 0.50 3 0.50 0.250 3 0.250 2 24 2700 26 1.188 0.6 365 7 7.4 1 4 : 35
ES, echo spacing; FoV, field of view; NSA, number of signal averages; PDw-TSE, proton density-weighted turbo spin echo; TE, echo time; TR, repetition time; TSE, turbo spin echo; wfs, water-fat shift.
birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
6 of 7
Table 2 Visibility and delineation of different anatomical structures of the temporomandibular joint at 1.5 and 3.0 T
Anatomical structure Disc Visibility anterior band Visibility intermediate zone Visibility posterior band Delineation Bilaminar zone Visibility Delineation Mandibular fossa Visibility Delineation Mandibular condyle Visibility Delineation Pterygoid muscle Visibility Delineation Overall image quality
1.5 T Reader 1 Mean SD
Reader 2 Mean SD
Interrater reliability Kappa p-value
3.0 T Reader 1 Mean SD
Reader 2 Mean SD
Interrater reliability Kappa p-value
2.75 2.75 2.83 2.37
0.53 0.67 0.63 0.49
2.79 2.79 2.87 2.33
0.58 0.65 0.61 0.48
0.91 0.78 0.77 0.72
,0.001 ,0.001 ,0.001 ,0.001
2.00 2.12 2.12 2.20
0.65 0.79 0.79 0.72
2.04 2.20 2.12 2.25
0.69 0.77 0.79 0.67
0.92 0.87 0.87 0.93
,0.001 ,0.001 ,0.001 ,0.001
2.79 2.70
0.65 0.62
2.75 2.70
0.60 0.55
0.92 0.84
,0.001 ,0.001
2.30 2.39
0.82 1.07
2.30 2.47
0.82 1.08
0.86 0.88
,0.001 ,0.001
2.25 2.25
0.60 0.44
2.20 2.29
0.58 0.46
0.76 0.89
,0.001 ,0.001
2.04 2.00
0.62 0.59
2.04 2.04
0.55 0.55
0.83 0.73
,0.001 ,0.001
2.20 2.41
0.65 0.77
2.12 2.45
0.74 0.77
0.86 0.93
,0.001 ,0.001
2.04 2.16
0.69 0.70
2.00 2.08
0.65 0.65
0.78 0.86
,0.001 ,0.001
2.75 2.41 2.62
0.44 0.50 0.49
2.79 2.45 2.58
0.41 0.50 0.50
0.64 0.74 0.73
0.001 ,0.001 ,0.001
1.87 1.83 1.95
0.74 0.48 0.55
1.87 1.91 1.91
0.68 0.58 0.50
0.85 0.81 0.90
,0.001 ,0.001 ,0.001
SD, standard deviation. Mean and SD are given for each structure. Grading was based on Likert scales ranging from 1 (excellent visibility) to 5 (not visible). For interrater reliability, kappa values and corresponding p-values are given.
surface coils are widely available. (c) Furthermore, the imaging protocol applied in the present study was specifically optimized to increase the visibility of all relevant structures of the TMJ at 3.0 T. In particular, the gains in SNR due to the higher magnetic field strength were used to increase the spatial resolution, yielding an additional improvement of the visibility of anatomical structures when compared with previous reported findings.13 (d) To the best of our knowledge, the present study was the first
Table 3 Between-group differences of visibility and delineation of anatomical structures of the temporomandibular joint at 1.5 and 3.0 T
Anatomical structure Disc Visibility anterior band Visibility intermediate zone Visibility posterior band Delineation Bilaminar zone Visibility Delineation Mandibular fossa Visibility Delineation Mandibular condyle Visibility Delineation Pterygoid muscle Visibility Delineation Overall image quality
1.5 T compared with 3.0 T Reader 1 Reader 2 Z-value p-value Z-value p-value 23.10 22.74 23.54 20.96
0.002 0.006 0.000 0.334
23.21 22.63 23.26 20.53
0.001 0.008 0.001 0.593
22.20 21.19
0.027 0.232
22.13 20.97
0.033 0.332
21.14 21.42
0.251 0.153
20.94 21.51
0.346 0.130
21.00 20.92
0.317 0.357
20.77 21.56
0.439 0.118
23.50 23.11 23.11
,0.001 0.002 0.002
23.57 22.96 23.23
,0.001 0.003 0.001
Wilcoxon signed-rank tests were performed for each structure. Z-values and corresponding p-values are given for each reader. Values in italics indicate that tests remained significant after correction for multiple comparisons (n 5 13).
Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
to underpin the aforementioned results by assessing and evaluating the SNR obtained with different sequences and field strengths to ensure a systematic reliability of the reported results. Limitations Several limitations have to be acknowledged. First, the sample size was small. The present study included only 12 asymptomatic healthy volunteers. However, the size of the present study group is in line with a previously published study.13 Second, only asymptomatic healthy volunteers were included. Although it is expected that the main findings of the present study may apply to patients as well, it has to be taken into consideration that TMDs are very heterogeneous, reaching from subtle soft-tissue lesions to clearly visible dislocations. Since the present study did not investigate the possible impact of the presented approach on the sensitivity and specificity of the MR-based diagnostic process in patients with TMDs, further studies are necessary to evaluate potential benefits of imaging the TMJ at 3.0 T with respect to various pathologies underlying TMDs. Third, MRI of the TMJ is frequently performed in closed- as well as in open-mouth position, increasing the diagnostic accuracy regarding displacement of the disc, whereas in the present study, imaging was performed in closed-mouth position only. However, since all assessed structures are perfectly visible in closed-mouth position, we suppose that the acquired images are sufficient to infer on potential benefits of imaging the TMJ at higher field strengths. Fourth, qualitative assessment of obtained MR images was performed only once by each reader without evaluating the intrareader agreement. However, interreader agreement between the two readers ranged from “substantial agreement” to “almost perfect agreement” for both field strengths, suggesting
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
7 of 7
robust and reliable reproducibility regarding the evaluated data sets. Fifth, there was lack of a gold standard. However, Sanal et al20 demonstrated recently that MRI of cadaveric TMJs at 3.0 T provides an excellent characterization of the TMJ’s anatomy, suggesting that MRI of the TMJ might represent an excellent surrogate for anatomical structures when performed in study volunteers.
which can easily be translated into clinical settings, thus potentially improving the diagnostic accuracy regarding possible pathologies underlying TMD.
Clinical implications The present study indicated that imaging the TMJ at 3.0 T yields a great potential for the clinical routine. First, imaging at 3.0 T allowed not only for a better evaluation of the temporomandibular disc but also of the masticatory muscles, likely facilitating an improved evaluation of the masticatory system in general. Second, 3.0-T MR scanners are broadly available at most MR sites. Thus, the expansion of the technique could soon accelerate. Finally, the presented optimized MR protocol and imaging set-up can easily be translated into clinical settings, since only commercially available equipment was used. Taken together, the present study presented an approach,
Imaging the TMJ at 3.0 T yielded significantly better image quality as well as visibility and delineation of clinically relevant anatomical structures than at 1.5 T using a commercially available two-channel surface coil and might therefore be preferred in clinical routine.
Conclusions
Conflict of interest This was a prospective institutional review board-approved MRI study. It included imaging of a phantom and a cohort of asymptomatic healthy volunteers. All volunteers gave written informed consent.
References 1. McNeill C. Management of temporomandibular disorders: concepts and controversies. J Prosthet Dent 1997; 77: 510–22. doi: 10.1016/S0022-3913(97)70145-8 2. Weinberg LA. The etiology, diagnosis, and treatment of TMJ dysfunction-pain syndrome. Part I: Etiology. J Prosthet Dent 1979; 42: 654–64. doi: 10.1016/0022-3913(79)90197-5 3. Maixner W, Diatchenko L, Dubner R, Fillingim RB, Greenspan JD, Knott C, et al. Orofacial pain prospective evaluation and risk assessment study–the OPPERA study. J Pain 2011; 12: T4–11.e12. doi: 10.1016/j.jpain.2011.08.002 4. Slade GD, Fillingim RB, Sanders AE, Bair E, Greenspan JD, Ohrbach R, et al. Summary of findings from the OPPERA prospective cohort study of incidence of first-onset temporomandibular disorder: implications and future directions. J Pain 2013; 14: T116–24. doi: 10.1016/j.jpain.2013.09.010 5. De Rossi SS. Orofacial pain: a primer. Dent Clin North Am 2013; 57: 383–92. doi: 10.1016/j.cden.2013.04.001 6. Hoffmann RG, Kotchen JM, Kotchen TA, Cowley T, Dasgupta M, Cowley AW Jr. Temporomandibular disorders and associated clinical comorbidities. Clin J Pain 2011; 27: 268–74. doi: 10.1097/ AJP.0b013e31820215f5 7. White BA, Williams LA, Leben JR. Health care utilization and cost among health maintenance organization members with temporomandibular disorders. J Orofac Pain 2001; 15: 158–69. 8. Manfredini D, Guarda-Nardini L, Winocur E, Piccotti F, Ahlberg J, Lobbezoo F. Research diagnostic criteria for temporomandibular disorders: a systematic review of axis I epidemiologic findings. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011; 112: 453–62. doi: 10.1016/j.tripleo.2011.04.021 ¨ 9. O˘gutcen-Toller M, Taskaya-Yilmaz N, Yilmaz F. The evaluation of temporomandibular joint disc position in TMJ disorders using MRI. Int J Oral Maxillofac Surg 2002; 31: 603–7. 10. Aiken A, Bouloux G, Hudgins P. MR imaging of the temporomandibular joint. Magn Reson Imaging Clin N Am 2012; 20: 397–412. doi: 10.1016/j.mric.2012.05.002 11. Iwasaki H, Kubo H, Harada M, Nishitani H. Temporomandibular joint and 3.0 T pseudodynamic magnetic resonance imaging. Part 1: evaluation of condylar and disc dysfunction. Dentomaxillofac Radiol 2010; 39: 475–85. doi: 10.1259/dmfr/29741224
12. Nordmeyer-Massner JA, Wyss M, Andreisek G, Pruessmann KP, Hodler J. In vitro and in vivo comparison of wrist MR imaging at 3.0 and 7.0 tesla using a gradient echo sequence and identical eight-channel coil array designs. J Magn Reson Imaging 2011; 33: 661–7. doi: 10.1002/jmri.22419 13. Stehling C, Vieth V, Bachmann R, Nassenstein I, Kugel H, Kooijman H, et al. High-resolution magnetic resonance imaging of the temporomandibular joint: image quality at 1.5 and 3.0 Tesla in volunteers. Invest Radiol 2007; 42: 428–34. doi: 10.1097/ 01.rli.0000262081.23997.6b 14. Schmid-Schwap M, Drahanowsky W, Bristela M, Kundi M, Piehslinger E, Robinson S. Diagnosis of temporomandibular dysfunction syndrome–image quality at 1.5 and 3.0 Tesla magnetic resonance imaging. Eur Radiol 2009; 19: 1239–45. doi: 10.1007/s00330-008-1264-7 15. Manoliu A, Spinner G, Wyss M, Ettlin DA, Nanz D, Kuhn FP, et al. Magnetic resonance imaging of the temporomandibular joint at 7.0 T using high-permittivity dielectric pads: a feasibility study. Invest Radiol 2015. Epub ahead of print. doi: 10.1097/ RLI.0000000000000196 16. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med 1990; 16: 192–225. doi: 10.1002/mrm.1910160203 17. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med 1999; 42: 952–62. doi: 10.1002/(SICI)1522-2594(199911)42: 5,952::AID-MRM16.3.0.CO;2-S 18. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–74. doi: 10.2307/2529310 19. Koh KJ, List T, Petersson A, Rohlin M. Relationship between clinical and magnetic resonance imaging diagnoses and findings in degenerative and inflammatory temporomandibular joint diseases: a systematic literature review. J Orofac Pain 2009; 23: 123–39. 20. Sanal HT, Bae WC, Pauli C, Du J, Statum S, Znamirowski R, et al. Magnetic resonance imaging of the temporomandibular joint disc: feasibility of novel quantitative magnetic resonance evaluation using histologic and biomechanical reference standards. J Orofac Pain 2011; 25: 345–53.
birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
RESEARCH ARTICLE
Quantitative and qualitative comparison of MR imaging of the temporomandibular joint at 1.5 and 3.0 T using an optimized high-resolution protocol 1,2 1
Andrei Manoliu, 2Georg Spinner, 2Michael Wyss, 3Stefan Erni, 3Dominik A Ettlin, 1Daniel Nanz, Erika J Ulbrich, 3Luigi M Gallo and 1Gustav Andreisek
1 Department of Radiology, Institute for Diagnostic and Interventional Radiology, University Hospital Zurich, University of Zurich, Zurich, Switzerland; 2Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zurich, Switzerland; 3 Center of Dental Medicine of the University of Zurich, Zurich, Switzerland
Objectives: To quantitatively and qualitatively compare MRI of the temporomandibular joint (TMJ) using an optimized high-resolution protocol at 3.0 T and a clinical standard protocol at 1.5 T. Methods: A phantom and 12 asymptomatic volunteers were MR imaged using a 2-channel surface coil (standard TMJ coil) at 1.5 and 3.0 T (Philips Achieva and Philips Ingenia, respectively; Philips Healthcare, Best, Netherlands). Imaging protocol consisted of coronal and oblique sagittal proton density-weighted turbo spin echo sequences. For quantitative evaluation, a spherical phantom was imaged. Signal-to-noise ratio (SNR) maps were calculated on a voxelwise basis. For qualitative evaluation, all volunteers underwent MRI of the TMJ with the jaw in closed position. Two readers independently assessed visibility and delineation of anatomical structures of the TMJ and overall image quality on a 5-point Likert scale. Quantitative and qualitative measurements were compared between field strengths. Results: The quantitative analysis showed similar SNR for the high-resolution protocol at 3.0 T compared with the clinical protocol at 1.5 T. The qualitative analysis showed significantly better visibility and delineation of clinically relevant anatomical structures of the TMJ, including the TMJ disc and pterygoid muscle as well as better overall image quality at 3.0 T than at 1.5 T. Conclusions: The presented results indicate that expected gains in SNR at 3.0 T can be used to increase the spatial resolution when imaging the TMJ, which translates into increased visibility and delineation of anatomical structures of the TMJ. Therefore, imaging at 3.0 T should be preferred over 1.5 T for imaging the TMJ. Dentomaxillofacial Radiology (2016) 45, 20150240. doi: 10.1259/dmfr.20150240 Cite this article as: Manoliu A, Spinner G, Wyss M, Erni S, Ettlin DA, Nanz D, et al. Quantitative and qualitative comparison of MR imaging of the temporomandibular joint at 1.5 and 3.0 T using an optimized high-resolution protocol. Dentomaxillofac Radiol 2016; 45: 20150240. Keywords: temporomandibular joint; temporomandibular joint disc; temporomandibular joint disorders; magnetic resonance imaging (MRI); signal-to-noise ratio (SNR)
Introduction Temporomandibular disorders (TMDs) are a collective term for various pathologies of the temporomandibular Correspondence to: Andrei Manoliu. E-mail: [email protected] The study was funded by the Swiss National Science Foundation grant number 320030_156466. Received 18 July 2015; revised 27 August 2015; accepted 9 September 2015
joint (TMJ) characterized by mutual features, such as pain, clicking or crepitus of the TMJ and alterations of the mouth opening path.1,2 With a prevalence of 12% in the adult population,3 TMDs are the most common cause for orofacial pain of non-dental origin.4 TMDs have a great impact on the quality of life of patients and
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
2 of 7
are related to a wide diversity of frequent disorders, including tension and migraine headache, depression, fatigue, sleep apnoea, obesity and Type 2 diabetes mellitus.5,6 In addition, TMDs cause great socioeconomic costs, causing 17,800,000 lost workdays for every 100,000,000 working adults per year in the USA.3,7 While the need for early detection of the pathological processes underlying TMD is evident, the diagnostic process is still subject of substantial discussion.8 In cases where imaging is required, MRI has been considered the imaging method of choice over the last decade to evaluate key features of TMD, such as temporomandibular disc displacement or alterations of cartilage and bones.9,10 Currently, MRI of the TMJ is mainly performed at 1.5 T using dedicated surface coils.11 Since the signal intensity [or, in particular, the signal-to-noise ratio (SNR)] is assumed to scale linearly with the field strength, theoretical considerations suggest that imaging the TMJ at higher field strengths such as 3.0 T should allow considerably higher spatial resolutions at comparable scan durations and therefore enable superior visibility of TMJ’s anatomical key structures.12 Although theoretical considerations suggest great advantages for MRI of the TMJ at higher magnetic field strengths, evidence in the literature for the use of 3.0 T is still scarce and only few studies13,14 investigated the benefit of imaging the TMJ at 3.0 T compared with 1.5 T. Stehling et al13 showed a better visibility of some anatomical structures, and Schmid-Schwap et al14 found a better diagnostic accuracy in patients with anterior disc displacement. However, these early studies had several limitations and only used commercially nonavailable custom-made bilateral phased-array surface coils or applied imaging protocols, which have not been optimized for imaging the TMJ at 3.0 T.13,14 In addition, the two aforementioned studies focused on qualitative assessment only, whereas the SNR and its potential impact on imaging results have not been quantitatively assessed so far. Therefore, clear evidence for a superior quantitative and qualitative performance of imaging the TMJ at 3.0 T using commercially available coils and easily reproducible dedicated imaging protocols is still missing and needs to be addressed. The aim of the present study was to quantitatively and qualitatively compare an optimized high-resolution protocol at 3.0 T and a clinical standard protocol at 1.5 T for imaging the TMJ using a commercially available two-channel surface coil. Methods and materials This was a prospective institutional review boardapproved MRI study. It included imaging of a phantom and a cohort of asymptomatic healthy volunteers. All volunteers gave written informed consent. The study was registered in the official research database of the University of Zurich. Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
Phantom imaging The first part of the study included standardized SNR measurements for which a “Braino” phantom (GE Medical Systems, Milwaukee, WI) containing 12.5 mM of N-acetyl-L-aspartic acid (NAA), 10 mM of creatine hydrate (Cr), 3 mM of choline chloride (Ch), 7.5 mM of myoinositol (mI), 12.5 mM of L-glutamic acid (Glu), 5 mM of DL-lactic acid (Lac), sodium azide (0.1%), 50 mM of potassium phosphate monobasic (KH2PO4), 56 mM of sodium hydroxide (NaOH) and 1 ml l21 of Gd-DPTA (Magnevist®; Bayer Healthcare Germany, Leverkusen, Germany) was used. The phantom was a model for the head of a human subject and was placed accordingly on the scanner’s patient table (Figure 1). Volunteer imaging For the second part of the study, 12 asymptomatic healthy volunteers (six females; mean age, 23.16 years; age range, 19–26 years; and six males; mean age, 25.33 years; age range, 20–29 years) who met the following inclusion criteria were included consecutively: age between 18 and 40 years and no past or current TMD. Exclusion criteria were pregnancy, metallic implants (including retainer) or any other contraindications for MRI (e.g. claustrophobia), as well as pain or functional symptoms in the area of the TMJ. All volunteers underwent simultaneous bilateral MRI of their TMJs with the jaw in closed position using the two-channel SENSE Flex S coils (Philips Healthcare, Best, Netherlands) of both scanners (Figure 2). All subjects were positioned in the supine and head-first position. Scans were performed consecutively. To avoid potential imaging bias, half of the volunteers were first assessed at 1.5 T and the other half were assessed at 3.0 T first. The assignment was made randomly. One of the investigators (blinded, AM) was present at all imaging sessions to ensure the correct positioning of the human subjects and coils. Both TMJs were imaged at the same time. MRI MRI was performed on a clinical 1.5-T Philips Achieva system and a clinical 3.0-T Philips Ingenia system (Philips Healthcare) using standard commercially available two-channel surface coils (SENSE Flex S; Philips Healthcare). The phantom and human subjects were imaged using the same hardware and software as well as the same imaging set-up. In general, sequences from the clinical standard protocol were applied. This included proton densityweighted turbo spin echo (PDw-TSE) MR sequences in oblique sagittal plane (field of view) positioned perpendicular to the mandibular condyle’s transverse axis as well as a PDw-TSE MR sequence in coronal plane. For imaging at 1.5 T, sequence parameters applied in clinical routine were used (Table 1). For imaging at 3.0 T, the PDw-TSE sequences were optimized by improving the in-plane resolution of the images (Table 1). Care was taken to keep the acquisition time similar and to not extend the scan duration. This study design was chosen
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
to reflect the clinical situation where 3.0-T protocols are typically different from 1.5 T to take advantage of the higher field strength. To allow standardized SNR measurements in the phantom part of the study, the clinical sequence was accompanied by an additional identical scan without radiofrequency excitation and without gradient switching. Image analysis Quantitative signal-to-noise ratio measurements in the phantom: According to a previously described method12 and strictly following a recently presented analysis algorithm,15 the SNR was examined on a voxelwise basis employing a so-called noise scan, which is basically an additional sequence where the radiofrequency excitation and gradients are switched off by applying the following equation using dedicated in-house software routines (MATLAB®; MathWorks®, Natick, MA): SNR5
jrj s
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dH c 2 1d
where r is the exact magnitude of the available transverse magnetization for each voxel according to Roemer et al,16 s is the standard deviation of its corresponding noise components as described by Pruessmann et al,17 d represents the complex values that a given pixel exhibits in each single-channel image, H denotes the complex conjugate transpose and c denotes the noise covariance matrix calculated from the noise scans as previously described.12,17 Therefore, the applied software routine accounted for possible noise correlations among all coil channels. Subsequently, the phantom was manually segmented by one author (GS), and SNR values were extracted, resulting in one SNR value for each field strength. Qualitative image evaluation in human subjects: All images were transferred to the University Hospital of Zurich’s picture archiving and communication system (IMPAX® 6.0; Agfa Healthcare, Mortsel, Belgium) and independently evaluated by two fellowship-trained radiologists (GA and AM; 14 and 4 years’ experience, respectively, with MRI of the musculoskeletal system). Both readers were blinded with regard to the subject’s details as well as field strength. In accordance to a previously reported grading system,13 images were evaluated in a random order for the overall image quality as well as the visibility and delineation of clinically relevant structures, i.e. the (a) TMJ disc (anterior band, intermediate zone, posterior band), (b) the bilaminar zone, (c) the mandibular fossa, (d) the mandibular condyle and (e) the pterygoid muscle on a 5-point Likert scale from 1 to 5 [1, excellent visibility; 2, good visibility; 3, moderate visibility; 4, poor visibility; and 5, not visible (complete lack of diagnostic information)]. Statistical analyses Two-sample t-tests were used to evaluate potential SNR differences between the images assessed in the phantom
3 of 7
at 1.5 and 3.0 T. Kappa statistics were used to determine the interreader agreement in the qualitative MR image analysis of human subjects. According to Landis and Koch,18 kappa values of 0.41–0.60 were considered as moderate agreement, values of 0.61–0.80 were considered as substantial agreement, values of 0.81–0.99 were considered as almost perfect agreement and values of 1.00 were considered as perfect agreement. To investigate statistically significant between-group differences regarding the visibility and delineation of clinically relevant anatomical structures as well as regarding the overall image quality using different field strengths, Wilcoxon signedrank tests were performed [a-level of 0.05, corrected for multiple comparisons (n 5 13)]. SPSS® software v. 22.0 (IBM Corporation, Armonk, NY; formerly SPSS Inc., Chicago, IL) was used for all statistical analyses. Results Quantitative signal-to-noise ratio data from phantom imaging Quantitative analysis revealed similar SNR for the optimized protocol at 3.0 T compared with the clinical standard protocol at 1.5 T without statistically significant differences: mean ± standard deviation: 1.5 T, 60.42 ± 9.28; 3.0 T, 56.48 ± 7.62; t 5 1.604; p 5 0.116 (Figure 1). Qualitative data from volunteer imaging 24 TMJ image sets of 12 healthy volunteers were evaluated. Interrater reliability ranged from “substantial agreement” to “almost perfect agreement” (Cohen’s kappa: 1.5 T, 0.647–0.934; 3.0 T, 0.739–0.932) for the different anatomical structures (Table 2). For both readers, qualitative analysis demonstrated significantly better overall image quality at 3.0 T than at 1.5 T (Wilcoxon signed-rank test: Reader 1, p 5 0.002; Reader 2, p 5 0.001). Furthermore, imaging at 3.0 T yielded significantly better visibility and delineation of the anterioposterior band of the temporomandibular disc as well as better visibility and delineation of the pterygoid muscle for both readers (Wilcoxon signed-rank tests, p , 0.05, corrected for multiple comparisons, Table 3 and Figure 2). In addition, imaging at 3.0 T yielded better visibility of the intermediate zone of the temporomandibular disc and better visibility of the bilaminar zone; however, statistical significance did not survive correction for multiple comparisons (Table 3). Discussion To the best of our knowledge, the present study demonstrated for the first time that imaging the TMJ at 3.0 T at high resolution yielded better quantitative and qualitative performance than at 1.5 T when using commercially available TMJ surface coils. Despite higher resolution at 3.0 T, SNR was similar, with both magnetic field strengths, suggesting that gains in SNR due to a higher magnetic field birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
4 of 7
Figure 1 Quantitative analysis. Panel (a) shows the spherical phantom. Two stacks consisting of 12 slices each were imaged in sagittal orientation resulting in 24 slices numbered consecutively from left to right. Panel (b) shows a diagram displaying the signal-to-noise ratio (SNR) for both 1.5 and 3.0 T in the spherical phantom. The x-axis represents the number of slices and the y-axis represents the SNR. Values for 1.5 T are given in red and values for the 3.0 T are given in blue. For each slice, the mean SNR is given for both field strengths. Panels (c) and (d) show the voxelwise distribution of SNR for each slice, respectively, whereas panel (c) represents 1.5 T and panel (d) 3.0 T. In both panels (c, d), SNR values are colourcoded from 0 (blue) to 120 (yellow). For colour image see online.
strength can be used to increase spatial resolution while maintaining a robust SNR. Quantitative analysis To the best of our knowledge, no study so far has investigated the SNR at different field strengths for imaging the TMJ. In the present study, we applied an intricate voxelwise approach to calculate SNR for different field strengths in a phantom.15 This approach yielded a comprehensive assessment of the spatial distribution of the SNR (Figure 1) with respect to the desired regions of assessment. Furthermore, our method yielded SNR maps for the whole field of view and correctly accounted for Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
possible noise correlations among all coil channels. Applying this method, we found that SNR was similar for the high-resolution protocol at 3.0 T and the clinical standard protocol at 1.5 T. This result is well in line with the fact that measured SNR is not only related to the magnetic field strength but also to the voxel size.12 In particular, SNR is negatively correlated with the spatial resolution; a higher spatial resolution (i.e. a smaller voxel size) is assumed to lower the SNR. In the present study, the gain in SNR due to the increased field strength at 3.0 T was used to increase the spatial resolution, yielding statistically similar SNR maps for both field strengths while the spatial resolution at 3.0 T was considerably higher
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
Figure 2 Qualitative analysis. proton density-weighted turbo spin echo oblique sagittal images in closed-mouth position at 3.0 T. Panels (a) and (b) show the images of a temporomandibular joint (TMJ) of an asymptomatic volunteer assessed at 1.5 T [panel (a), “good” overall image quality] and 3.0 T [panel (b), “excellent” image quality] using a dedicated, commercially available TMJ surface coil (FlexS; Philips Healthcare). In general, the overall image quality was significantly better on images obtained at 3.0 T. In particular, the visibility and delineation of the anterior band and the posterior band is increased at 3.0 T. See Table 3 for detailed presentation of between-group differences regarding visibility and delineation of all anatomical structures of the TMJ.
than at 1.5 T. Particularly when considering that in many cases the correlation between reported clinical symptoms and MRI findings in patients with TMDs is still unsatisfying,19 exploiting SNR gains at higher field strengths to increase the spatial resolution should be desirable from a clinical perspective, since a better and more detailed depiction of the TMJ’s anatomical subregions is assumed to improve diagnostic sensitivity when evaluating potential pathologies underlying TMDs. Qualitative analysis Using a high-resolution protocol, imaging the TMJ at 3.0 T yielded better visibility and delineation for several key structures of the TMJ, including the temporomandibular disc and the pterygoid muscle. Furthermore, overall image quality was higher at 3.0 T than at 1.5 T. Statistical differences remained significant after correction for multiple
5 of 7
comparisons. It is to note, however, that the depiction of the temporomandibular fossa and condyle did not benefit from increased field strength, which is however clinically less important, since they can be already well depicted at 1.5 T. These findings are in line with previous preliminary studies,13,14 reporting better visibility of the temporomandibular disc at 3.0 T than at 1.5 T. In particular, Stehling et al13 reported increased visibility of the temporomandibular disc in asymptomatic volunteers when imaged at 3.0 T compared with 1.5 T. However, a dedicated custommade phased-array surface coil was used for imaging at 3.0 T, whereas a commercially available surface coil was used at 1.5 T. In another study, Schmidt-Schwap et al14 found that imaging the TMJ at 3.0 T yielded better visibility of the position and shape of the disc and therefore provides increased diagnostic accuracy compared with 1.5 T in patients with suspected anterior disc displacement. However, the sequence parameters at both field strengths were kept almost identical, thus not exploiting potential benefits of higher magnetic field strengths, such as increased spatial resolution. Furthermore, both studies13,14 focused on the temporomandibular disc and condyle only, neglecting other anatomical structures of high relevance regarding TMDs, such as the pterygoid muscle. Therefore, our findings extend the current knowledge as follows. (a) In addition to better visibility and delineation of the temporomandibular disc, we found better visibility and delineation of the pterygoid muscle, particularly the part near the condylar insertion. These findings are of essential clinical interest, since the pathophysiology of TMDs is highly complex and can only be accurately assessed when evaluating the complete musculoskeletal functional unit of the TMJ, including the masticatory muscles. (b) In contrast to previous studies,13 only commercially available coils (FlexS; Philips Healthcare) were used for imaging at both field strengths, ensuring that the reported results are not explained by different array designs. Moreover, the current results can be translated directly into a clinical setting, since the used
Table 1 Scan parameters of the PDw sequences at 1.5 and 3.0 T
Parameter FoV (mm) Pixel size (mm) Reconstructed pixel size (mm) Slice thickness (mm) Number of slices TR TE Wfs in pixel (mm) Effective wfs in image (mm) Wfs (Hz) TSE factor ES NSA Scan time (min)
PDw-TSE sagittal 1.5 T 150 3 150 0.60 3 0.75 0.375 3 0.375 2 2 3 12 2000 30 0.937 0.6 231 7 7.5 2 03 : 48
3.0 T 150 3 150 0.50 3 0.50 0.250 3 0.250 2 2 3 12 2700 26 1.199 0.6 362 7 7.4 1 03 : 52
PDw-TSE coronal 1.5 T 120 3 180 0.60 3 0.75 0.375 3 0.375 2 24 2000 30 0.99 0.6 217 8 8 2 4 : 04
3.0 T 120 3 180 0.50 3 0.50 0.250 3 0.250 2 24 2700 26 1.188 0.6 365 7 7.4 1 4 : 35
ES, echo spacing; FoV, field of view; NSA, number of signal averages; PDw-TSE, proton density-weighted turbo spin echo; TE, echo time; TR, repetition time; TSE, turbo spin echo; wfs, water-fat shift.
birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
6 of 7
Table 2 Visibility and delineation of different anatomical structures of the temporomandibular joint at 1.5 and 3.0 T
Anatomical structure Disc Visibility anterior band Visibility intermediate zone Visibility posterior band Delineation Bilaminar zone Visibility Delineation Mandibular fossa Visibility Delineation Mandibular condyle Visibility Delineation Pterygoid muscle Visibility Delineation Overall image quality
1.5 T Reader 1 Mean SD
Reader 2 Mean SD
Interrater reliability Kappa p-value
3.0 T Reader 1 Mean SD
Reader 2 Mean SD
Interrater reliability Kappa p-value
2.75 2.75 2.83 2.37
0.53 0.67 0.63 0.49
2.79 2.79 2.87 2.33
0.58 0.65 0.61 0.48
0.91 0.78 0.77 0.72
,0.001 ,0.001 ,0.001 ,0.001
2.00 2.12 2.12 2.20
0.65 0.79 0.79 0.72
2.04 2.20 2.12 2.25
0.69 0.77 0.79 0.67
0.92 0.87 0.87 0.93
,0.001 ,0.001 ,0.001 ,0.001
2.79 2.70
0.65 0.62
2.75 2.70
0.60 0.55
0.92 0.84
,0.001 ,0.001
2.30 2.39
0.82 1.07
2.30 2.47
0.82 1.08
0.86 0.88
,0.001 ,0.001
2.25 2.25
0.60 0.44
2.20 2.29
0.58 0.46
0.76 0.89
,0.001 ,0.001
2.04 2.00
0.62 0.59
2.04 2.04
0.55 0.55
0.83 0.73
,0.001 ,0.001
2.20 2.41
0.65 0.77
2.12 2.45
0.74 0.77
0.86 0.93
,0.001 ,0.001
2.04 2.16
0.69 0.70
2.00 2.08
0.65 0.65
0.78 0.86
,0.001 ,0.001
2.75 2.41 2.62
0.44 0.50 0.49
2.79 2.45 2.58
0.41 0.50 0.50
0.64 0.74 0.73
0.001 ,0.001 ,0.001
1.87 1.83 1.95
0.74 0.48 0.55
1.87 1.91 1.91
0.68 0.58 0.50
0.85 0.81 0.90
,0.001 ,0.001 ,0.001
SD, standard deviation. Mean and SD are given for each structure. Grading was based on Likert scales ranging from 1 (excellent visibility) to 5 (not visible). For interrater reliability, kappa values and corresponding p-values are given.
surface coils are widely available. (c) Furthermore, the imaging protocol applied in the present study was specifically optimized to increase the visibility of all relevant structures of the TMJ at 3.0 T. In particular, the gains in SNR due to the higher magnetic field strength were used to increase the spatial resolution, yielding an additional improvement of the visibility of anatomical structures when compared with previous reported findings.13 (d) To the best of our knowledge, the present study was the first
Table 3 Between-group differences of visibility and delineation of anatomical structures of the temporomandibular joint at 1.5 and 3.0 T
Anatomical structure Disc Visibility anterior band Visibility intermediate zone Visibility posterior band Delineation Bilaminar zone Visibility Delineation Mandibular fossa Visibility Delineation Mandibular condyle Visibility Delineation Pterygoid muscle Visibility Delineation Overall image quality
1.5 T compared with 3.0 T Reader 1 Reader 2 Z-value p-value Z-value p-value 23.10 22.74 23.54 20.96
0.002 0.006 0.000 0.334
23.21 22.63 23.26 20.53
0.001 0.008 0.001 0.593
22.20 21.19
0.027 0.232
22.13 20.97
0.033 0.332
21.14 21.42
0.251 0.153
20.94 21.51
0.346 0.130
21.00 20.92
0.317 0.357
20.77 21.56
0.439 0.118
23.50 23.11 23.11
,0.001 0.002 0.002
23.57 22.96 23.23
,0.001 0.003 0.001
Wilcoxon signed-rank tests were performed for each structure. Z-values and corresponding p-values are given for each reader. Values in italics indicate that tests remained significant after correction for multiple comparisons (n 5 13).
Dentomaxillofac Radiol, 45, 20150240
birpublications.org/dmfr
to underpin the aforementioned results by assessing and evaluating the SNR obtained with different sequences and field strengths to ensure a systematic reliability of the reported results. Limitations Several limitations have to be acknowledged. First, the sample size was small. The present study included only 12 asymptomatic healthy volunteers. However, the size of the present study group is in line with a previously published study.13 Second, only asymptomatic healthy volunteers were included. Although it is expected that the main findings of the present study may apply to patients as well, it has to be taken into consideration that TMDs are very heterogeneous, reaching from subtle soft-tissue lesions to clearly visible dislocations. Since the present study did not investigate the possible impact of the presented approach on the sensitivity and specificity of the MR-based diagnostic process in patients with TMDs, further studies are necessary to evaluate potential benefits of imaging the TMJ at 3.0 T with respect to various pathologies underlying TMDs. Third, MRI of the TMJ is frequently performed in closed- as well as in open-mouth position, increasing the diagnostic accuracy regarding displacement of the disc, whereas in the present study, imaging was performed in closed-mouth position only. However, since all assessed structures are perfectly visible in closed-mouth position, we suppose that the acquired images are sufficient to infer on potential benefits of imaging the TMJ at higher field strengths. Fourth, qualitative assessment of obtained MR images was performed only once by each reader without evaluating the intrareader agreement. However, interreader agreement between the two readers ranged from “substantial agreement” to “almost perfect agreement” for both field strengths, suggesting
MRI of the temporomandibular joint at 1.5 and 3.0 T Manoliu et al
7 of 7
robust and reliable reproducibility regarding the evaluated data sets. Fifth, there was lack of a gold standard. However, Sanal et al20 demonstrated recently that MRI of cadaveric TMJs at 3.0 T provides an excellent characterization of the TMJ’s anatomy, suggesting that MRI of the TMJ might represent an excellent surrogate for anatomical structures when performed in study volunteers.
which can easily be translated into clinical settings, thus potentially improving the diagnostic accuracy regarding possible pathologies underlying TMD.
Clinical implications The present study indicated that imaging the TMJ at 3.0 T yields a great potential for the clinical routine. First, imaging at 3.0 T allowed not only for a better evaluation of the temporomandibular disc but also of the masticatory muscles, likely facilitating an improved evaluation of the masticatory system in general. Second, 3.0-T MR scanners are broadly available at most MR sites. Thus, the expansion of the technique could soon accelerate. Finally, the presented optimized MR protocol and imaging set-up can easily be translated into clinical settings, since only commercially available equipment was used. Taken together, the present study presented an approach,
Imaging the TMJ at 3.0 T yielded significantly better image quality as well as visibility and delineation of clinically relevant anatomical structures than at 1.5 T using a commercially available two-channel surface coil and might therefore be preferred in clinical routine.
Conclusions
Conflict of interest This was a prospective institutional review board-approved MRI study. It included imaging of a phantom and a cohort of asymptomatic healthy volunteers. All volunteers gave written informed consent.
References 1. McNeill C. Management of temporomandibular disorders: concepts and controversies. J Prosthet Dent 1997; 77: 510–22. doi: 10.1016/S0022-3913(97)70145-8 2. Weinberg LA. The etiology, diagnosis, and treatment of TMJ dysfunction-pain syndrome. Part I: Etiology. J Prosthet Dent 1979; 42: 654–64. doi: 10.1016/0022-3913(79)90197-5 3. Maixner W, Diatchenko L, Dubner R, Fillingim RB, Greenspan JD, Knott C, et al. Orofacial pain prospective evaluation and risk assessment study–the OPPERA study. J Pain 2011; 12: T4–11.e12. doi: 10.1016/j.jpain.2011.08.002 4. Slade GD, Fillingim RB, Sanders AE, Bair E, Greenspan JD, Ohrbach R, et al. Summary of findings from the OPPERA prospective cohort study of incidence of first-onset temporomandibular disorder: implications and future directions. J Pain 2013; 14: T116–24. doi: 10.1016/j.jpain.2013.09.010 5. De Rossi SS. Orofacial pain: a primer. Dent Clin North Am 2013; 57: 383–92. doi: 10.1016/j.cden.2013.04.001 6. Hoffmann RG, Kotchen JM, Kotchen TA, Cowley T, Dasgupta M, Cowley AW Jr. Temporomandibular disorders and associated clinical comorbidities. Clin J Pain 2011; 27: 268–74. doi: 10.1097/ AJP.0b013e31820215f5 7. White BA, Williams LA, Leben JR. Health care utilization and cost among health maintenance organization members with temporomandibular disorders. J Orofac Pain 2001; 15: 158–69. 8. Manfredini D, Guarda-Nardini L, Winocur E, Piccotti F, Ahlberg J, Lobbezoo F. Research diagnostic criteria for temporomandibular disorders: a systematic review of axis I epidemiologic findings. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011; 112: 453–62. doi: 10.1016/j.tripleo.2011.04.021 ¨ 9. O˘gutcen-Toller M, Taskaya-Yilmaz N, Yilmaz F. The evaluation of temporomandibular joint disc position in TMJ disorders using MRI. Int J Oral Maxillofac Surg 2002; 31: 603–7. 10. Aiken A, Bouloux G, Hudgins P. MR imaging of the temporomandibular joint. Magn Reson Imaging Clin N Am 2012; 20: 397–412. doi: 10.1016/j.mric.2012.05.002 11. Iwasaki H, Kubo H, Harada M, Nishitani H. Temporomandibular joint and 3.0 T pseudodynamic magnetic resonance imaging. Part 1: evaluation of condylar and disc dysfunction. Dentomaxillofac Radiol 2010; 39: 475–85. doi: 10.1259/dmfr/29741224
12. Nordmeyer-Massner JA, Wyss M, Andreisek G, Pruessmann KP, Hodler J. In vitro and in vivo comparison of wrist MR imaging at 3.0 and 7.0 tesla using a gradient echo sequence and identical eight-channel coil array designs. J Magn Reson Imaging 2011; 33: 661–7. doi: 10.1002/jmri.22419 13. Stehling C, Vieth V, Bachmann R, Nassenstein I, Kugel H, Kooijman H, et al. High-resolution magnetic resonance imaging of the temporomandibular joint: image quality at 1.5 and 3.0 Tesla in volunteers. Invest Radiol 2007; 42: 428–34. doi: 10.1097/ 01.rli.0000262081.23997.6b 14. Schmid-Schwap M, Drahanowsky W, Bristela M, Kundi M, Piehslinger E, Robinson S. Diagnosis of temporomandibular dysfunction syndrome–image quality at 1.5 and 3.0 Tesla magnetic resonance imaging. Eur Radiol 2009; 19: 1239–45. doi: 10.1007/s00330-008-1264-7 15. Manoliu A, Spinner G, Wyss M, Ettlin DA, Nanz D, Kuhn FP, et al. Magnetic resonance imaging of the temporomandibular joint at 7.0 T using high-permittivity dielectric pads: a feasibility study. Invest Radiol 2015. Epub ahead of print. doi: 10.1097/ RLI.0000000000000196 16. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med 1990; 16: 192–225. doi: 10.1002/mrm.1910160203 17. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med 1999; 42: 952–62. doi: 10.1002/(SICI)1522-2594(199911)42: 5,952::AID-MRM16.3.0.CO;2-S 18. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–74. doi: 10.2307/2529310 19. Koh KJ, List T, Petersson A, Rohlin M. Relationship between clinical and magnetic resonance imaging diagnoses and findings in degenerative and inflammatory temporomandibular joint diseases: a systematic literature review. J Orofac Pain 2009; 23: 123–39. 20. Sanal HT, Bae WC, Pauli C, Du J, Statum S, Znamirowski R, et al. Magnetic resonance imaging of the temporomandibular joint disc: feasibility of novel quantitative magnetic resonance evaluation using histologic and biomechanical reference standards. J Orofac Pain 2011; 25: 345–53.
birpublications.org/dmfr
Dentomaxillofac Radiol, 45, 20150240
